基于Haar—like矩形特征的交通标志识别

针对道路交通标志识别系统中的识别精度和识别实时性问题，提出一种基于Haar—like矩形特征的交通标志识别算法。为了提高提取速度，用积分图像先对样本库进行处理，提取Haar-like矩形特征做为特征向量。提取的特征向量因维数过高．应用PCA降维法对数据进行降维。将降维后的特征向量用来对支持向量机训练。实验结果表明，Hart-like矩形特征向量具有形式简单．计算速度快等优点。在小样本的目标识别中，利用SVM进行分类识别，取得很好的识别效果．并在识别耗时上有所优化。

Traffic Sign Recognition Based on Harr-like Rectangle Features

Proposes an algorithm based on Haar-like rectangle features of the traffic signs to improve the instantaneity and accuracy in real-time traffic sign recognition. In order to increase the speed of extraction, uses the integral image to deal with the original images and extract Haar-like rectangle features as the characteristics of the data. Because the characteristics of the data extracted is too much, applies the PCA technology to reduce the dimensions of the data, which can appropriately remove some part of the feature vector that have less in- formation. Uses the reduced characteristics of the dates to train the SVM. Experimental results indicate that Haar-like rectangle features has simple form and fast calculation. The algorithm shows better accuracy and has less time consumption with SVM in traffic sign recogni- tion.